01Backlog
Requests from Canny and our own ideas used to sit apart, ranked by separate formulas. I merged them into one list under our rules, so customer demand is now a live part of every decision instead of a slide we check once a quarter.
How we evolved
Our feature ideas lived in a Google Sheet, where we weighed them to decide which came next. I’d circulate it, gather feedback, and get sign-off. Meanwhile, requests from clients and employees poured into Canny, and its rigid formula ranked them differently from ours. Once priorities were settled, we estimated how many weeks of work a project would require from PMs, designers and developers. Then I recreated the plan in Asana, faking sprints with milestones and shuffling dates until nobody was overworked. That makeshift board was our roadmap.
Today it happens in one place. Everything, Canny’s included, sits in the same list, scored our way. It also understands capacity: it counts heads per team, leaves a buffer for surprises, and knows when a PM must step in on design. We size up projects exactly as before, but the math is automatic, and anyone overbooked is flagged up front. Live progress flows in from Jira, and Claude writes a short summary for every category, ready to send to everyone.
We retired the spreadsheet and the Asana workaround, and since it stays current, we’re no longer chased for updates. We start a new quarter knowing what we can actually deliver.
For the teams that rely on it
Requests from Canny and our own ideas used to sit apart, ranked by separate formulas. I merged them into one list under our rules, so customer demand is now a live part of every decision instead of a slide we check once a quarter.
Every project’s size comes straight from the backlog and is weighed against how much the team can take on each quarter. The app flags when we go over, so we can shift work around before committing. Claude summarizes each theme, making the plan easy to explain.
Matching projects to people used to mean math by hand and shuffling in Asana. Now the app counts heads, leaves room for surprises and plans for PMs covering design, so we see who’s overbooked before a sprint begins.
Behind the scenes
I had never built anything with AI. I made the whole app with it, from scratch, in one week.
It followed the scoring and staffing methods we already trusted, so nobody had to learn a new way of working.
Requests from Intercom chats and employees flowed in from Canny, and only the important ones surfaced.
Live progress came straight from Jira into a simple view, so anyone could check where a project stood without logging in there or asking a PM.
What changed